Today, Americans over 65 can expect to spend 10-50% of their remaining lives in a dependent state. Prolong active life expectancy (i.e., decreasing the absolute years of dependency as people survive to older ages) is thus an urgent goal. This is particularly critical for Mexican Americans (MA), who compris3e the most rapidly growing segment of the U.S. elderly population and appear to have markedly higher rates of dependence than European Americans (EA). Prolonging active life expectancy will require systematic and detailed information about stages in the progression from specific diseases to ultimate dependence, and about factors which modify these stages, in order to identify optimal targets for effective interventions to slow or prevent this progression in MA as well as EA elderly. We therefore assembled a unique, aging, bi-ethnic cohort of community- dwelling MA and EA elderly of low, middle, and high socioeconomic status. An epidemiological analysis of this cohort indicated that ethnic differences in impairments, functional limitations, and disability may result primarily from the higher prevalence of diabetes among MA elderly. Further multi-variable, hierarchical analyses, using the Displacement Process Model (DPM) as a framework, enabled us to identify 1) a psychosocial, and lifestyle modifiers to variables within that main pathway. These cross-sectional results suggest relationships that might indicate promising targets for intervention, but distinguishing the valid targets among these will require a longitudinal study to estimate the time course of these relationships within the DPM. This proposed further study is designed: 1) to define optimal targets for intervention in the main disease-to-disability pathways of diabetes and arthritis; and 2) to determine whether psychosocial and lifestyle factors alter the main pathways of these diseases and regardless of disease, whether psychosocial and lifestyle factors influence successive stages in the main DPM pathway, suggesting that these factors themselves are promising targets for function-enhanced intervention. To estimate the time course of relationships in the DPM, we will carry out three follow-up assessments of the cohort at 18-month intervals. Longitudinal growth curve models will be used to fit person-specific curves to these longitudinal data, and estimate the general shapes of time- related changes among the cohort. This new study will provide critical information identifying key points in the disablement process where targeted medical and/or behavioral interventions may be most effective in altering the course of disability. Thus, scientific and public health benefits will be substantial.